The reference group for all categories is individuals with no opioids dispensed to family members. The odds ratios of overdose and 95% CIs were not reported for individuals aged 60 years and older in the categories of opioid dispensing to family members of 50 to less than 90 morphine milligram equivalents (MMEs) and 90 or more MMEs per day because of unstable estimates.
eFigure 1. Matched case-control study design for an exposed case and exposed control
eTable 1. Opioid drugs list.
eTable 2. ICD-9-CM code definitions of comorbidities.
eTable 3. Other drug dispensings.
eTable 4. Immediate-release/short-acting (IR/SA) non-steroidal anti-inflammatory drugs.
eTable 5. Extended-release/long-acting (ER/LA) non-steroidal anti-inflammatory drugs.
eTable 6. ICD-9-CM code definitions of pain diagnoses.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Khan NF, Bateman BT, Landon JE, Gagne JJ. Association of Opioid Overdose With Opioid Prescriptions to Family Members. JAMA Intern Med. 2019;179(9):1186–1192. doi:10.1001/jamainternmed.2019.1064
Are opioid prescriptions to family members associated with overdose among individuals who themselves do not have an opioid prescription?
In this case-control study, prior opioid dispensing to family members was associated with 2.89-fold higher odds of individual overdose, which persisted in young children and increased with greater quantities of opioid medications dispensed to family members.
Family member prescriptions may be a risk factor for overdose.
Prescription opioid misuse is a public health problem that leads to overdose. Although existing interventions focus on limiting prescribing to patients at high risk, individuals may still access prescription opioids dispensed to family members.
To determine whether opioid prescriptions to family members were associated with overdose for individuals who themselves did not have an opioid prescription.
Design, Setting, and Participants
We conducted a 1:4 matched case-control study using health care utilization data from 2004 through 2015 from a large US commercial insurance company. Eligible individuals were required to have at least 12 months of continuous enrollment and 1 or more family members in the database. Individuals who experienced overdose were identified by their first opioid overdose after the baseline period and matched to control participants by time in the database, calendar time, age, sex, and number of individuals in the family unit. Both groups were restricted to individuals with no prior opioid dispensing of their own. Data analysis was conducted from January 2018 to August 2018.
Any prior opioid dispensing to a family member, total morphine milligram equivalents dispensed to family members, and the type of opioid product dispensed.
Main Outcomes and Measures
Individual odds of opioid overdose resulting in an emergency department visit or hospitalization were the primary end point. The primary analysis evaluated the odds of overdose among individuals whose family members had been dispensed an opioid. Sensitivity analyses examined the odds stratified by age and timing relative to the dispensing of opioids to family members.
A total of 2303 individuals who experienced opioid overdose and 9212 matched control individuals were identified. The mean (SD) age was 23.2 (18.1) years; 1158 affected individuals and 4632 control individuals (50.3%) were female. The mean (SD) time in the database before an overdose case was 3.2 (3.3) years. Prior opioid dispensing to family members was associated with individual overdose (odds ratio [OR], 2.89 [95% CI, 2.59-3.23]). There was a significant dose-response association between increasing amounts of opioids dispensed to family members and odds of overdose (>0-<50 morphine milligram equivalents per day: OR, 2.71 [95% CI, 2.42-3.03]; 50-<90 morphine milligram equivalents per day: OR, 7.80 [95% CI, 3.63-16.78]; ≥90 morphine milligram equivalents per day: OR, 15.08 [95% CI, 8.66-26.27]).
Conclusions and Relevance
In this analysis, opioid prescriptions to family members were associated with overdose among individuals who do not receive opioid prescriptions. Interventions may focus on expanding access to opioid antagonists, locking prescription opioids in the home, and providing greater patient education to limit fatal overdose among family members.
The US Centers for Disease Control and Prevention estimated that opioid overdose was responsible for more than 42 000 deaths in the United States in 2016 alone.1 In that year, approximately 11.5 million people misused prescription opioid medications,2 and more than 2 million individuals misused prescription opioid medications for the first time.3 In October 2017, the US Department of Health and Human Services declared a public health emergency in response to the ongoing opioid epidemic.4 Data from state departments of health suggest that the rates of overdose mortality have continued to increase.5,6
In an attempt to curb opioid misuse and overdoses, several strategies to reduce individual opioid exposure have been implemented. Comprehensive prescription drug monitoring programs that track state-level prescribing and dispensing of controlled substances and allow prescribers to query the database to identify whether a patient is at risk prior to writing an opioid prescription have been found to reduce opioid dosages.7 A number of states have also implemented or are considering laws that impose limits on the daily supply or dosage of opioid prescriptions that can legally be prescribed.8 While these approaches aim to limit the number of opioid prescriptions that an individual receives or the amount of opioids received per prescription, individuals may have access to other sources of opioids, such as from prescriptions to family members.
Results of the Monitoring the Future Study and the National Survey on Drug Use and Health indicate that most individuals who misuse opioids obtain them from family members.9,10 Up to 90% of patients report having unused opioids after surgery,11-13 and residing in the same household as someone with an opioid prescription has been associated with an increased likelihood of subsequently obtaining one’s own opioid prescription.14,15 A small case-control study in Ontario, Canada, found that children 10 years or younger whose mothers were prescribed opioids were 2.4 times more likely to present to the hospital with an opioid overdose than children whose mothers were not prescribed opioids.16 Several other studies have identified a leftover supply of prescription opioids in the household as a major contributor to overdose in children, in that nearly all accidental ingestion in childhood is attributable to family members’ prescriptions.17-19
The objectives of this study were to examine whether opioid prescriptions to family members are associated with overdose among individuals who themselves do not have opioid prescriptions. We also sought to define how this risk varies by age, the quantity of opioids dispensed, and the timing relative to the dispensing of opioids to family members.
We conducted this nested case-control study with a large national administrative claims database. We used data spanning January 1, 2004, to September 30, 2015, from Optum Clinformatics Data Mart. The end date was selected to precede the transition to the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) system for outcome classification in the claims data. The database contains deidentified administrative pharmacy and medical claims for a large population across all 50 US states who were insured by a national commercial insurer. The data also contain an encrypted subscriber number that identifies subscribers and their dependents who are covered under the same health insurance plan.
This study was approved by the Institutional Review Board at Brigham and Women’s Hospital. Informed consent was not required because the database contained only deidentified information to protect patient confidentiality.
We established a cohort of individuals with at least 365 days of continuous enrollment in the database and at least 1 other individual (or family member) insured under the same health plan. Affected individuals (with cases of opioid overdose) were identified by their first opioid overdose in the database after this baseline period and could not have had any prior incidents of opioid dispensing or overdoses. For each case, 4 control individuals20 were selected from among cohort members still under follow-up at the affected individual’s overdose date (on the scale of time since the end of the baseline period) and who did not have any prior incidents of opioid dispensing or overdoses in the database up until and including the overdose date. Because opioid overdose is uncommon in individuals who do not have their own prescriptions for opioids, control individuals were matched to affected individuals on time in the database, calendar time, age, sex, and number of individuals on the same health plan to increase statistical efficiency.21,22
For all affected and control individuals, we assessed occasions in which opioid prescriptions were dispensed to family members during follow-up (ie, the window between the end of the 12-month baseline period and the matched overdose date). We also summed the total morphine milligram equivalents (MME)23 for all opioids dispensed to all family members of each affected individual and control individual during follow-up and further estimated the mean (SD) MMEs dispensed per day to family members (eFigure in the Supplement). We assessed whether opioid prescriptions to family members were for extended-release/long-acting products or fentanyl patches (eTable 1 in the Supplement).
Overdose was defined as the first emergency department visit or hospitalization associated with opioid overdose, based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. We defined definite affected individuals using codes for opioid-associated poisoning (965.00, 965.02, and 965.09) or accidental poisoning (E850.1 and E850.2). Individuals with probable cases were defined using codes for adverse effects of opioid use (E935.1 and E935.2) and at least 1 code for overdose-associated symptoms, including acute respiratory failure, hypoxia, or apnea (codes 518.81, 518.82, 786.03, 786.05, 786.09, 799.0*, and 799.1*) on the same day.24 Codes used to define definite overdose, together with codes for poisoning by heroin, have been found to have a specificity of 99.9%.25
In addition to the variables used for matching, we assessed a range of covariates during the baseline period. These ranged from use of other drugs, including sedative hypnotics, antidepressants, and other nonopioid analgesics; to diagnosis of opioid dependence; diagnosis of substance use (of alcohol, tobacco, marijuana, cocaine); diagnosis of drug addiction or associated diagnoses (HIV, hepatitis B, and hepatitis C); other comorbidities, including depression, bipolar disorder, and schizophrenia; and diagnoses of pain conditions (eTables 2-6 in the Supplement).
We used conditional logistic regression to estimate odds ratios (ORs) and 95% CIs for the association between opioid overdose and whether individuals had family members with a prior opioid dispensing. We conditioned the model on the matched sets to control for confounding owing to the matching factors. We further adjusted for all other measured covariates as independent variables in the model. In dose-response analyses, we categorized total dosage of opioids provided to family members as greater than 0 to less than 50 MMEs per day, 50 or more to less than 90 MMEs per day, and 90 or more MMEs per day, using no opioid prescriptions as the referent.26 We also conducted stratified analyses among children, adolescents, and adults, creating groups defined by ages 0 to 6 years, 7 to 12 years, 13 to 18 years, 19 to 34 years, 35 to 59 years, and 60 years or older.
We conducted 2 prespecified sensitivity analyses. We first assessed opioid dispensing to family members restricted to the 30 and 90 days immediately preceding the overdose to examine whether more recent opioid prescriptions to family members have a larger association with odds of overdose. We also restricted the case definition to definite cases of opioid overdose. To contextualize the OR estimates, we estimated the baseline rate of overdose in the absence of opioid prescriptions among individuals with at least 365 days of continuous enrollment and at least 1 family member in the database who themselves did not have any opioid prescriptions and had no family members with any opioid prescriptions in the claims data. Analyses were performed using SAS version 9.4 (SAS Institute). All P values were 2-sided, and the level of significance was 0.05. Data analysis occurred from January 2018 to August 2018.
We identified 2303 individuals affected by overdose (of whom 2204 [95.7%] had definite cases of overdose) and 9212 matched control individuals (Figure 1). The mean (SD) age was 23.2 (18.1) years, 1158 affected individuals and 4632 control individuals (50.3%) were female, and most (1610 affected individuals and 6440 control individuals [69.9%]) shared the same health plan with 4 or fewer other individuals. In the absence of opioid prescriptions to individuals and family members, the baseline rate of overdose was 3.80 (95% CI, 3.55-4.05) events per 100 000 person-years among individuals with 365 days of continuous enrollment and at least 1 family member in the database. The mean (SD) time in the database before an overdose occurred was 3.2 (3.3) years. Baseline characteristics of matched individuals affected by overdose and control individuals are presented in Table 1. As expected, affected individuals and control individuals had different distributions of baseline characteristics. After adjusting for these differences in the conditional logistic regression analyses, any prior opioid dispensing to family members was associated with individual overdose (OR, 2.89 [95% CI, 2.59-3.23]). This result was similar when restricted to individuals with definite cases of opioid overdose (OR, 3.01 [95% CI, 2.69-3.37]). The OR for the association between overdose and dispensing to family members of extended-release/long-acting opioid prescriptions was 2.92 (95% CI, 2.43-3.51) and 4.31 (95% CI, 2.46-7.54) for fentanyl patches (Table 2).
We also observed a dose-response association between the amount of opioids dispensed to family members and the odds of individual overdose. After adjustment for baseline covariates, dispensing opioids to family members in amounts greater than 0 to less than 50 MMEs per day was associated with 2.71 (95% CI, 2.42-3.03) times the odds of overdose. The OR increased to 7.80 (95% CI, 3.63-16.78) when family members were dispensed between 50 and less than 90 MMEs per day and 15.08 (95% CI, 8.66-26.27) with 90 or more MMEs dispensed per day.
In age-stratified analyses, any prior dispensing of opioids to family members was associated with a 4.08-fold (95% CI, 3.07-5.41) increase in the odds of overdose in children age 0 to 6 years (Figure 2). The ORs were 4.38 (95% CI, 3.07-6.26) in children age 7 to 12 years, 3.38 (95% CI, 2.78-4.11) in adolescents age 13 to 18 years, 1.78 (95% CI, 1.33-2.38) in young adults age 19 to 34 years, 2.28 (95% CI, 1.77-2.96) in adults age 35 to 59 years, and 1.80 (95% CI, 0.96-3.37) in adults 60 years and older. Dose-response associations were observed within age groups (OR ranges as dosages increased: ages 0-6 years, 3.64-34.22; 7-12 years, 4.24-21.71; 13-18 years, 3.27-8.38; 19-34 years, 1.71-10.10; 35-59 years, 2.02-56.25).
When the window for assessing exposure was restricted to the 90 days before the index date, the ORs were comparable (≥90 MMEs per day: 90-day exposure: OR, 15.66 [95% CI, 9.11-26.94] vs full period: OR, 15.08 [95% CI, 8.66-26.27]) or higher (for >0-<50 MMEs/d: 90-day exposure: OR, 3.63 [95% CI, 3.11-4.24]; full period: OR, 2.71 [95% CI, 2.42-3.03]; 50-<90 MME per day: 90-day exposure: OR, 13.49 [95% CI, 6.17-29.49 vs full period: OR, 7.80 [95% CI, 3.63-16.78]) than the primary analysis results that assessed exposure during all of follow-up (Table 3). When restricted to 30 days before the index date, the ORs further increased for all exposure variables (for >0-<50 MMEs/d: 30-day exposure: OR, 4.14 [95% CI, 3.37-5.09]; full period: OR, 2.71 [95% CI, 2.42-3.03]; 50-<90 MME per day: 30-day exposure: OR, 11.40 [95% CI, 4.70-27.69 vs full period: OR, 7.80 [95% CI, 3.63-16.78]), except for the highest-dosage category (≥90 MMEs per day: 30-day exposure: OR, 15.44 [95% CI, 9.40-25.35]; full period: OR, 15.08 [95% CI, 8.66-26.27]). (Odds ratios were not estimated in the categories >50-<90 MMEs per day and ≥90 MMEs per day for those ≥60 years because there was only 1 exposed case in these groups combined.)
In this nested case-control study within a large commercially insured population in the Unites States, we found that opioid overdose among individuals who themselves do not have opioid prescriptions is rare, but that the rate increases nearly 3-fold in association with the dispensing of opioids to family members. This risk was present across all age groups, among children, adolescents, and adults. We observed an apparent dose-response association based on the quantity of opioid medications dispensed to family members and stronger associations when focused on family member prescriptions occurring in the 30 or 90 days prior to overdose. The dispensing of extended-release/long-acting opioids and fentanyl patches to family members were also associated with a higher likelihood of overdose. These results suggest that opioid prescriptions to family members may be an important risk factor for overdose.
These results are similar to those from a previous small, unadjusted nested case-control study (n = 103 affected individuals) that assessed the risk of overdose among young children whose mothers were prescribed opioids. The study used prescription claims data from the Ontario Public Drug Benefit Database and found that children 10 years or younger who overdosed were more likely to have a mother who received an opioid prescription.16 This study corroborates these findings in a larger commercially insured US population with substantially more incidents of overdose. We were also able to evaluate overdose risk stratified by age, opioid prescription dosage, and type of opioid dispensed. The database used included an encrypted subscriber number, which allowed us to identify all family members insured under the same health plan over long periods of follow-up. Thus, we could assess the risk of overdose associated with the dispensing of opioids to family members among more family members than mothers and children and evaluate overdose risk in both children and adults.
The finding of a consistently elevated association across age groups is interesting because the mechanism of overdose is likely different for young children than adolescents and adults. Overdose among young children is likely a result of accidental exposure to opioids or other substances of abuse. Intentional misuse may be more likely among adolescents and adults, to the extent that opioid prescriptions to family members are implicated in these overdoses. It is important to note that, although the ORs between the dispensing of opioids to family members and overdose are larger for children than adults, these relative measures may not correspond to similar differences on the absolute scale since the overall rate of overdose is much smaller among young children.
Although we cannot determine whether access to family members’ opioids led to the higher overdose rate or whether individuals with family members who receive opioid prescriptions are more likely to overdose for other reasons, these findings have important implications for current overdose prevention strategies to address the ongoing opioid epidemic. Since access to family members’ prescription opioids appears to be either a strong risk factor or a marker for overdose, interventions may focus on increasing the use of secure medication storage boxes in the home. Patient education is also integral to reducing accidental exposure and misuse, because many patients may not be aware that their own opioid prescriptions may put their family members at risk of overdose. Effective communication by physicians, pharmacists, nurses, or public service announcements could increase awareness of opioids as a risk factor for family member overdose, advise patients not to share their medications with others even if they have pain issues, and educate patients and family members of the availability of opioid antagonists to treat overdose. Non-patient-specific prescriptions allow pharmacy or lay distribution of opioid antagonists to individuals who themselves do not have a prescription for an opioid antagonist or an opioid.
This study has several limitations. The data end date of September 30, 2015, used to precede the ICD-10-CM transition, may not reflect current patterns in opioid prescribing and overdose events. Although the encrypted subscriber number allowed the assessment of family-level exposures, there may be other individuals in the household who are not insured under the same health plan. As a result, there may be additional opioid prescriptions that could not be included in the exposure assessment. Alternatively, individuals under the same health plan may reside in a different household from their family members. Although we adjusted for a large number of patient characteristics, this study was not oriented to assess if the association between opioid prescriptions to family members and individual overdose is causal. For instance, individuals with family members who were prescribed opioids could be more likely to access and overdose on illicit drugs, such as heroin or other opioids, obtained outside of the family unit. Nevertheless, the strength of the association implies that access to family members’ prescription opioids is an important factor associated with patients being at high risk of overdose.
Furthermore, the opioid overdose outcome definition in claims data does not distinguish between prescription opioid–associated and illicit opioid–associated overdose events. We attempted to account for this by adjusting for illicit drug use, but it is typically underreported in claims data. The data are also limited to overdoses that present to the emergency department and hospitals. Incidents of death that occur without the patient reaching the emergency department or hospital would not be identified, which may have reduced the number of affected individuals in this study. However, it is unlikely to have resulted in bias, given the high specificity (99.9%) of the outcome definition.
The results of this large, nationwide case-control study suggest that opioid dispensing to family members is a strong risk factor or marker for overdose among individuals who themselves do not have an opioid prescription. These findings persist in children and adolescents and are heightened in the short period immediately after family members are dispensed opioids and among family members to whom larger quantities of opioids are dispensed. Interventions may focus on opioid dispensing limits, encouraging patients to properly dispose of prescription opioids after use, enhancing patient and public education, and using secure medication storage boxes to limit the risk of opioid overdose among other household family members. Treating and preventing opioid overdose events at the family level, as well as expanding access to opioid antagonists, should also be considered.
Accepted for Publication: March 12, 2019.
Corresponding Author: Joshua J. Gagne, PharmD, ScD, Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, One Brigham Circle, Ste 3030, Boston, MA 02120 (email@example.com).
Published Online: June 24, 2019. doi:10.1001/jamainternmed.2019.1064
Author Contributions: Ms Khan and Dr Gagne had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Khan, Gagne.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Khan, Landon.
Critical revision of the manuscript for important intellectual content: Khan, Bateman, Gagne.
Statistical analysis: Khan, Landon.
Supervision: Bateman, Gagne.
Conflict of Interest Disclosures: Dr Bateman has received salary support from grants from Eli Lilly and Company, GlaxoSmithKline, Pacira, Baxalta, and Pfizer to the Brigham and Women’s Hospital for unrelated work, served as a consultant to Aetion Inc and on a project sponsored by Merck for Mothers, and received personal fees from Alosa Foundation. Dr Gagne has received salary support from grants from Eli Lilly and Company and Novartis Pharmaceuticals Corporation to the Brigham and Women’s Hospital and has been a consultant to Aetion Inc and Optum Inc outside of the submitted work. No other disclosures were reported.
Funding/Support: The study was funded internally by the Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital.
Role of the Funder/Sponsor: The Division of Pharmacoepidemiology and Pharmacoeconomics in the Department of Medicine at the Brigham and Women’s Hospital had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.